ED 366 509

AUTHORTITLE

INSTITUTION

SPONS AGENCY

PUB DATECONTRACTNOTE

AVAILABLE FROM

PUB TYPE

EDRS PRICEDESCRIPTORS

IDENTIFIERS

ABSTRACT

DOCUMENT RESUME

SE 054 136

Donmoyer, RobertRethinking the Form and Function of ScientificResearch in Science Education. NCSTL MonographSeries. #5.National Center for Science Teaching and Learning,Columbus, OH.Office of Educational Research and Improvement (ED),Washington, DC.Jul 92R117Q0006213p.; First presented at the Annual Meeting of theAmerican Educational Research Association (SanFrancisco, CA, 1992).National Center for Science Teaching and Learning,1929 Kenny Road, Columbus, OH 43210-1080,Reports Research/Technical (143)

MF01/PC01 Plus Postage.Classroom Research; Educational Anthropology;Educational Change; Educational History; *EducationalPolicy; *Educational Research; Elementary SecondaryEducation; Higher Education; Qualitative Research;*Science Education; Science TeachersDeweys Experimentalism; Journal of Research inScience Teaching; *National Center for ScienceTeaching and Learning; Reform Efforts; ScienceEducation Research; Thorndike (Edward L)

The basic premise explored in this paper is that theconception of educational research which has dominated thinking inthe past and continues to influence how the research community thinksand what it does today is inadequate and to some extent,inappropriate. This paper is divided into four sections. Section 1explicates the traditional conception of research in education anddiscusses the role research was traditionally expected to play ineducation. Section 2 focuses on more contemporary views about theform and function of research in education. Evidence of both growingskepticism about traditional views and continued acceptance oftraditional notions is presented. The third section focuses onproblems with the traditional views of research and the traditionalview of research's role in applied public policy fields such aseducation. The final part of the paper focusos on implications of thetwo problems discussed in Section 3 for rethinking the form andfunction of educational a research in general and for constructing aresearch agenda for the National Center for Science Teaching andLearning. (PR)

***********************************************************************

Reproductions supplied by EDRS are the best that can be madefrom the original document.

***********************************************************************

Rethinking the Formand Function of

Scientific Research inScience Education

Robert Donrnoyer

U.S. DEPARTMENT OF EDUCATIONOf lice of Educational Rematch end improvement

EDUCATIONAL RESOURCES INFORMATIONCENTER (ERIC)

Xl This document has beim reproduced asreceived from the person or organizationonginsting it

0 Minor changes here men made to improvereproduction quality

Points 01 vilive of opinions slated in thilidoeu-ment do not necessarily represent officialOE RI position or policy

NCSTL Monograph Series. # 5July 1992

Iii' \\IIAN\ICINTIR

SCIENCE1 1CHING&

LEAIRNINGRosearch Conks 104

1314 Kinnear RoadColumbus, OH 43212

(614) 292-3339

BEST CRY AVAILABLE

FIN \IJO\CI\IIR

SCIENC11it MI INC &11 MING

The NCSTL Monograph SeriesNumber 5

Rethinking the Form andFunction of Scientific Research inScience Education

Robert Donmoyer

In this review and critique of traditional research in science education, ProftssorRobert Donmoyer of The Ohio State Universiv argues that the tradtional mode ofresearch is inadequate and inappropriate, and that the problems of "idiayncraty andimmaculate perception"will require dramatic change within the research community

ifresearch is to have a positive impact on science education practice .

THE BASIC PREMISE TO BE EXPLORED HERE CAN BE

STATED SUCCINCILY: the conception of educationalresearch which has dominated our thinking in thepast and continues to influence how we think andwhat we do today is inadequate and, to some ex-tent, inappropriate. The paper is divided into foursections. Section 1 explicates the traditional con-ception of research in education and discusses therole research was traditionally expected to play inthe field. Section 2 focuses on more contemporaryviews about the form and function of research ineducation; evidence of both growing skepticismabout traditional views and continued acceptanceof traditional notions is presented.

This paper was first presented at the American Educational ResearchAssociation Annual Meeting, San Francisco, 1992. Writing ofthispaper war supported by the National Center firr Science Teachingand Learning under grant # R117Q00062 _from the Office ofEducational Research and Improvement, U.S. Department ofEducation. Any opinion:, finding4 conclusions, or recommendationsexpressed in this publication are those of the author and do notnecessari# relect the views of the sponsoring agencies.

3

The third section focuses on problems with thetraditional view of research and the traditionalview of research's role in applied public policyfields such as education. Two problems are dis-cussed: the problem of idiosyncracy and the prob-lem of immaculate perception. These problems, itis argued are sufficiently serious to require morethan modest, incremental adjustments in tradition-al ways of thinking about and traditional ways ofdoing research.

The final part of the paper focuses on ixnplica-tions of the two problems discussed i n Section 3 forrethinking the form and function of educationalresearch in general and for constructing a researchagenda for The National Center for Science Teach-ing and Learning in particular.

Traditional Notions

DURING THE FIRST HALF OF THIS CENTURY PROGLIES-

SIVE educators had two interrelated items on their

2

reform agenda: (1) taking the schools out of poli-tics and (2) establishing a cadre of professionals whothey assumed would make educational decisions onthe basis of research rather than political consid-erations. Implicit in the early progressives view ofboth professionalism and educational research wasa social engineering metaphor which emphasizedestablishing research based standard operatingprocedures for practice and hierarchical control toinsure these procedures got implemented. (Calla-han, 1964: Tyack, 1972).

Often the social engineering metaphor was evenmade explicit Franklin Bobbitt (1924), the fatherof the curriculum field, began his classic text, Howto Make a Curriculunz, by likening the work ofcurriculum developers to the work of eagineers.Similarly, the father of the educational adminis-tration field, Ellwood P. Cubberly, wrote in 1909:

Our schools are, in a sense, factories in whicli theraw products (children) are to be shaped an4fashioned into products to meet the various de-mands of life. The specifications for manufactur-ing come from the demands for thetwentieth-century civilization, and it is the busi-ness of the school to build its pupils according tothe specifications laid down. This demands goodtools, specialized machinery, continuous mea-surement of production to see if it is according tospecifications, the elimination of waste in manu-facture, and a large variety in the output (p.338).

Raymond Callahan's (1964) historical accountof the ir fluence of the efficiency movement onschool administration and historian David Tyack's(1974) description of school administrators' searchfor a "one best system" demonstrate that cubber-ly and Bobbitts' social engineering orientation wasshared by the field in general.

Researchers encouraged educators' socialengineering view of professionalism and suggestedthat research could provide the necessaryknowledge base to make social engineering possible.In 1910, for instance, Thorndike wrote in the leadarticle of the inaugural issue of The Journal of

Educational Psychology that

[a] complete science of psychology would tellery fact about everyone's intellect and characterand behavior, would tell t he cause of everychange in human nature, would tell the resultwhich every educational forceevery act of ev-ery person that changed any other or the agenthimselfwould have. It would aid us to use hu-man being for the world's welfare with the samesurety of the result that we now have when weuse falling bodies or chemical elements. In pro-portion as we get such a science we shall becomemasters of our own souls as we are now mastersof heat and light. Progress toward such a scienceis being made (p. 6).

John Dewey, of course, proposed a somewhatdifferent, much less mechanistic version of socialengineering. As a historian of American educationrecently noted, however, understanding the historyof American education in the Twentieth Centurybegins with the realization that Dewey lost andThorndike won.

Mechanistic visions of social engineering imag-es did not leave pn..,hic consciousness after earlyprogressives such as Thorndike, Cubberly andBobbitt were no longer players in the public are-na. As late as the 1970's federal policymakersfunded elaborate planned variation studies whichpolicy analysts assured them would tell which pol-icies and programs were most effective and conse-quently which should be mandated or funded.Within education, Project Follow Through (AbtAssociates, 1977: Haney & Villaume, 1977) is thebest known example of a planned variation study.The goal of this program was to determine the rel-ative effectiveness of different early childhoodeducation models in educating disadvantaged (theadjective of choice of the time) students.

Other indicators that social engineering imag-ery influenced thought and practice well into thesecond half of this century include the use of dis-crepancy models of program evaluation which as-sessed particular programs by comparing them toa model program which research established as

4

successful in another site; the majority of the re-search conducted in educational psychology includ-ing process-product studies of teaching, virtuallythe only research on teaching conducted prior tothe 1970's (Good, Biddle, & Brophy, 1975); a hostof programs and practices such as competencybased teacher education and competency based

"The question should not

be, Do they deviate?;...but rather, 'Are theyadapting well to theirrespective

environments?'"teacher evaluation which were legitimated by ei-ther explicit or implicit references to process-prod-uct studies of teacher effectiveness; and thepublication, in 1963, of Campbell and Stanley'sExperimental and Quasi-Experimental Designs forResearch, a book which codified Thorndike's visionof research and served (and continues to serve) as a rulebook for many who play the research game.

Contemporary Thinking

OVER THE PAST FIFTEEN OR SO YEARS ONE CAN FIND

evidence both of a growing skepticism about thetraditional vision of educational research and ofthe resilience of traditional views. This sectiondocuments both trends.

INCREASED SKEPTICISM. Over time, members ofthe research community, at least, have becameincreasingly skeptical about researchers' ability todeliver the sort of knowledge base which wouldmake social engineering possible. The inability ofthe Project Follow Through to provide definitiveresults, for instance, generated skept-ism in theminds of many who previously bad espoused thevirtues of planned variation studies (Rivlin & Tim-pane, 1975). One group of scholars, after review-

5

3

ing Project Follow Through data, took note of theprobabalistic nature of the findings of the studyand indicated that this aspect of the findings

should be honored widely and serve as a basis ofeducational policy. Local schools do seem tomake a difference. The peculiarities of individu-al teachers, schools, neighborhoods, and homesinfluence pupils' achievement far more thanwhatever is captured by labels such as basicRkills or affective education (House, Glass,McLean, & Walker, 1978, p. 462).

Discrepancy evaluation models also have begunto be criticized. Spindler, for example, wrote thesecomments about a discrepancy-oriented evaluationof programs established by the Youth EmploymentDemonstration Act: "My first reaction was, 'Whywould anyone expect different programs in differ-ent urban sites to replicate a model program inanother site?' This expectation is against the firstlaw of sociocultural systems in that all such systems(and a program of any kind is a sociocultural sys-tem) are adaptations to their environment. Weshould expect each program to show significantdeviation from an initiating model, and from eachof the other programs. The question should not be,'Do they deviate?' or even 'How do they deviate?',but rather, 'Are they adapting well (functionally)to their respective environments?'" (Spindler, ascited in Fetterman, 1981, p. 70).

Furthermore, even Thorndike's field of educa-tion psychology has undergone some rather dramat-ic changes over the past several decades. By themid 70's for instance, a new line of research onteachingone which focused on the complex pro-cess of teacher thinking rather than discrete teach-ing behaviors (e.g., Clark & Yinger, 1977; Shuhnan& Lz.nier, 1977) had begun to be established. EvenGage, a the-hard supporter of process-product mod-els of research on teaching, was forced to acknowl-edge in 1978 that such research could, at best, onlyprovide a f.r,eneral knowledge base for teaching andthat teacher artistry would always be required toadjust and shape that knowledge base to the needsof particular students and particular situations.

4

Today process-product studies are difficult tofind within research on teaching literature. Thatliterature is dominated by studies of teacher think-ing, a subject which is normally investigated withmethods more associated with the largely descrip-tive discipline of anthropology than withThorndike's social engineering oriented field ofstudy. Even single case studies, the sort of studieswhich Campbell and Stanley's rulebook indicatedhad "such a total absence of control as to be of al-most no scientific value" (Campbell and Stanley,1963, p. 6) have begun to be published by some ofour most prestigious and selective journals, includ-ing The Journal of Research in Science Teaching.

VESTIGES OF THE PAST. Playing counterpoint tothe evidence above is evidence which suggests thatsocial engineering imagery is alive and well andvery much present in the thinking of contemporarypolicymakers and even researchers. Evidence ofthe resilience of traditional modes of thinking isparticularly plentiful in the areas of policymakingand program development

Teacher proof curricula such as DISTAR, forinstance, are still being used in schools across thenation, especially in special education programs(Kuder, 1990). Also states such as Florida, Tennes-see, and Texas have used the probabalistic findingsof teacher effectiveness research to design check-list type instruments to assess teachers' competenceand make certification and merit pay decision.Florida did abolish its merit pay system recentlywhen some of the state's most dedicated and gift-ed teachers failed to score high enough on thestate's effectiveness instrument to qualify for ex-tra compensation, but elsewhere the practice ofusing fool-proof, research based systems to assessand reward teacher performance continues.

At the federal level, the latest buzzword is sys-temic reform. Both the National Science Founda-tion and the Office of Educational Research and:mprovement have endorsed the cgncept and haveattempted to reorganize at least some of their pro-grams around it. It is too early to know preciselyhow this concept will play out in practice (indeed,at the moment different federal agencies seem to

be attaching different meanings to the term), butthe rhetoric suggests, fairly unequivocally, thatsocial engineering imagery undergirds the notionof systemic reform. Furthermore, the rhetoric onsystemic reform seems match closer to the Cubber-ly/Thorndike mechanic dc version of social engi-neering than it does t J the less mechanistic, lesscontrol oriented version propounded by Dewey.

There are, of course, some significant differenc-es between earlier mechanistic views of social en-gineering and more contemporary systemic reforminitiatives. The NSF version of systemic reform,for instance relies less on social science research asa basis for decision making and more on politicalcoalition building. Also, most systemic reformmodels, at least, acknowledge the need for somediscretionary decision making at the local level.(See, for example, Smith, 1991.)

Similar vestiges of earlier ways of thinking canbe seen within the research community. Withinthe research community, however, vestiges of thepast seem less a conscious recommitment to thenotion of social engineering and more a somewhatthoughtless adherence to tradition and the stan-dard operating procedures of the past. This phe-nomenon is no more evident than in the field ofscience education. Indeed, a reliance on traditionand traditional ways of doing and thinking aboutresearch is displayed quite clearly on the pages ofThe Journal of Research in Science Teaching, thepremier research journal of that field within theUnited States.

Let me quickly acknowledge that The Journalof Research in Science Teaching has publishedsome truly innovative work in the last few years.This work, however, remains on the periphery ofthe field. Evidence of this fact can be found in aneditorial which appeared in a 1991 issue of thejournal. In the editorial, the journal's editor, RonGood, reprinted the guidelines which are sent toall reviewers of manuscripts which have been sub-mitted to the journal. After acknowledging thatthe guidelines "have been used by previous JRSTeditors in nearly the same form, before qualitativeresearch in science education became as prevalent

6

5

Figure 1: JRSTReviewer Guidelines

SOME GUIDELINES FOR REVIEWThe following items/questions are intended to assist you in evaluating 'aid writing a review of the enclosed manuscript. This list isnot necessarily exhaustive nor will each item apply to every type of manuscript. You will have to choose those CritCria which ap-ply to the manuscript enclosed and supplement the list as the need arises. You might wish to use your responses to the appropri-ate criteria in composing your review of the manuscript.

1. rdltafftraa. Is it descriptive of the study?b. Will it facilitate easy retrieval through sesrch system?c. Should the title be changed? If yes, what do you suggest?d. Does it contain key words or phrases needed in information search systems?2. &staga. Is it specific enough to communicate the principal parts of the paper?b. Ls it succinct and accurate? If not, which parts should be removed or changed?c. Is it missing any critical information? If so, what?3. Introduction/Rationalea. Does it discuss the importance of the study for science teaching?b. Does it provide a link between the problem and the study design?c. Does it establish a relationship between the study and previous work?d. Is the rationale based on pertinent, essential work or vague generalizations?e. Is the defmition of the problem adequate?1. Are the specific questions reasonable in light of previous research?4. Itelktiia. Is the method justified?b. Is the sample documented and properly selected?c. Are the treatments specified in sufficient detail to allow for replication?d. Are the models used documented and explained?e. Are techniques for data collection appropriate to the enquiry? Have they been adequately specified?1. Are data collection instruments valid and reliable? Are they justified?g. Is the statistical power of the study discussed?h. If tne study tests hypotheses statistically, are the safeguards against error explained and defended?5. Bra&a. Is the data analysis appropriate to the question?b. Is the data analysis sufficient? (Are means, standard deviations, sums of squares, degrees of freedom, explained variancere-ported where appcopriate?)c. Are the data tables easy to read and complete?d. Are the necessary data reported? If not, what is needed?e. Are unnecessary data reported? If so, what should be deleted?1. Are the illustrations appropriate/necessary?

g. Has an adequate description been given of the setting of the study and observations made to present a convincing case?6. lotczurtatinaa. Are the conclusions appropriate to the findings of the study?b. Are alternative interpretations recognized and discussed?c. Are limitations to the study identified and discussed (low power, multiple significance, tests, etc.)?d. Is the discussion congruent with the introduction/rationale for the study?e. Are the implications for science teaching specified and explained?f. Did the author(s) make an effort to translate theory into practice?7. Haw=a. Is the reference list adequate for the study?b. Are key references missing? If so, which ones?c. Are the refererres outdated, inaccurate?d. Are the teferences cited accurately used?8. aenCLiatallosa. Is the paper easy to follow?b. Are headings used appropriately?c. Should specific sections be shortened/lengthened?d. Is the writing style concise? Is the argument clear?

7

6

as it is today" (p. 291), Gond presents a list of thepresent actual guidelines (Figure 1).

What is interesting here is not just that manyof these criteria are quite inappropriate to assess-ing qualitative work (a problem Good acknowledg-es) or that it has taken so long to recognize this andbegin to do something about it (One of Good's ac-knowledged purposes for printing the guidelineswas to solicit suggestions about how to "make theguidelines more sensitive to both qualitative andquantitative research issues" 1p. 2911.). What isparticularly interesting about the editorial is thetotal absence of even one hint that the utility anddesirability of the guidelines listed and of the workproduced by those who follow the guidelines mayneed to be reconsidered and critically examined.Rather what we see is a response which smacks ofincrementalism. A few more people will be let inthe tent; an alternative entrance might even be pro-vided and a new wing mighteven be built. (This is thestrategy employed by theAmerican Educational Re-search Association. TheAmerican Educational Re-search Journal now has twodistinctly different sections,each with its own editor, dis-course style, and even refer-

tally. In the next section of this paper I will dis-cuss two problems which I believe call out for moreradical solutions.

Two Problems

I have applied labels to the two problems I wantto discuss. One I call the problem of idiosyncracy;the other I have dubbed the problem of immacu-late perception.

THE PROBLEM OF IDIOSYNCRACY. Let me try to il-

luminate the first problem by means of a story.The story is abo ui. two of my colleagues at OhioState. Several years ago these two colleagues re-ceived a research grant to study young children'sdevelopment. The two professors made a ratherodd couple. One was an accomplished social scien-tist by training and temperament, a person well

schooled in statistical

The response: "Teachers

do not teach cctegoriesor types; they teachchildren.

111

IIIM1111.

ence procedures.) Fundamental questions aboutthe form and function of the space we inhabit willbe avoided, however.

Incrementalism, of course, is highly function-al. Schema tbDorists, for example, note that it isnormal for individuals to try to assimilate noveltyinto existing ways of thinking and acting and, whennovelty cannot be assimilated, to try to accommo-date novelty without totally upsetting one's concep-tual applecart. These tendencies promotepsychologized stability, virtue, to be sure.

Furthermore, stability is at least as much of avirtue at the organizationallsociological level as itis at the level of the individuals and individualpsychology. There comes a time, however, whenthe problems are too great to be resolved incremen-

analysis and researchdesign. His co-inves-tigator was equallybright, but she hadspent most of her ca-reer working as ateacher. Their dif-ferent backgroundscreated creative ten-

sions and many disagreements.

One ongoing problem revolved around theteacher's complaint that none of the statisticaldescriptions they were generating described any ofthe actual students they had studied. The socialscientist acknowledged this fact but could not seewhy his co-investigator considered this a problem.Social science research, he explaincd patiently (andeventually somewhat impatiently) focused on com-monalities and generalities; it described types orcategories of people not actual people. To this theteacher/teacher educator's response was always thesame: Teachers do not teach categories or types;they teach children.

The teacher's comments get to the heart of theproblem of idiosyncracy. The problem can be stat-

8

ed as follows: social science is about ideal types; theconcern of a field like education, however, is ulti-mately individuals not aggregates.

The research community's growing realizationof this problem and its significance can be seen inthe career of the eminent educational psychologist,Lee Cronbach. By the mid 1950s Cronbach hadalready established himself as a skilled player ofthe Thorndike/Campbell and Stanley researchgame. In 1957, however, Cronbach told the Amer-ican Psychological Association that the complexi-ty of human phenomena requires a minoralteration in the traditional research game plan.Rather than searching for laws which were univer-sal and context free. Cronbach argued, research-ers should attempt to identify cause-effectrelationships between certain educational treat-ments on the one hand and certain types of indi-viduals (in Cronbach's terms, individuals withcertain aptitudes) on the other.

In the mid 1970s, however, after nearly twen-ty years of searching for aptitude x treatment in-teractions and nearly twenty years of frustrationbrought on by "inconsistent findings coming fromroughly similar inquiries," Cronbach (1975) toldthe American Psychological Association:

Once we attend to interactions, we enter a hall of''kirrors that extends to infinity. However far we

our analysisto third order or fifth orderor any otheruntested intetactions of still higherorder can be envisioned. (p. 119)

Compounding the problem of Complexity wasthe problem of culture. Cronbach cited Bronfen-brenner's historical look at child rearing practicesof middle- and lower-class parents. Class differenc-es documented in the 1950s were often just thereverse of practices that had been observed in the1930s. Cronbach concluded:

The trouble, as I see it, is that we cannotstore up generalizations and constructs forultimate assembly into a network. It is asif we needed a gross of dry cells to power

an engine and could only make one amonth. The energy would leak out of

7

the first cells before we had half thebattery compkted. So it is with thepotency of our generalizations (p.123).

In his 1975 article, Cronbach emphasized thatthe social world was no less lawful than the phys-ical world. The problem was that social laws weretoo complex and the social world too changeable toidentify them. By the early 1980s, however, Cron-bach had rejected even the notion of social laws. Hebegan suggesting that the entire cause-effect wayof thinking which undergirds the traditional viewof research is an inappropriate way to cHaracter-ize social phenomena. By 1982, in fact, Cronbachhad arrived at a position similar to that of symbolicinteractionists (Blumer, 1969) and ethnomethod-ologists (Garfinkle, 1967): Human action is con-structed not caused; those who expect research toproduce the sort of definitive cause-effect gener-alizations promised by Thorndike are simply, inCronbach's words, "Waiting for Godot."

Not everyone has arrived at as radical a posi-tion as Cronbach's, of course. Phillips (1987), forinstance, has argued that Cronbach has underes-timated the complexity of physical phenomenaand, hence, overestimated the relative complexi-ty of phenomena in the social world. Others mightpoint out that researchers are quite capable of gen-erating probabalistic findings which can, at least.inform us of the likelihood that a particular edu-cational treatment will produce a particular edu-cational outcome. Few, however, would dispute thefact that even if Phillips is correct when he arguesthat the social world is no more complex than thephysical world (a questionable assumption giventhe fact that Phillips does not even address thechangeability of culture issue raised by Cronbach),educational purposes are almost always more com-plex. An engineer employing the theory of quan-ttun mechanics, for example, is not interested inwhat happens to individual electrons; probahalis-tic findings, therefore, will be more than adequateto accomplish the engineer's purpose. Teachers,however, do care about individual students; prob-abalistic findings, therefore, have limited utilityin accomplishing complex educational purposes. Aprobabalistic finding, after all, not only tells us

8

what is likely to occur; it also reminds us that whatis unlikely may very well happen.

In short, no matter how large our sample, wecan never know whether research findings willapply to a new and different case or even to a par-ticular case in the original sample. This fact cer-tainly does not require that we totally reject thetraditional research game. It does require, jutwev-er, that we reject Thorndike's notion that socialscience research can provide the sort of knowledgebase which could support social engineering.

THE PROBLEM OF IMMACULATE PERCEPTION. The

problem of immaculate perception has been recog-nized in an array of disciplines and fields of study,although each discipline and field tends to charac-terize the problem in a somewhat different way.Philosophers of science talk in terms of paradigmsor conceptual frameworks. Psychologists use theterms schema or cognitive structures. Anthropolo-gists talk of cultural constructs; literary critics ofinterpretive frameworks. I am not suggesting thateach of these terms is a precise synonym for theother; each, however, alludes to the fact that we donot have direct access to reality and that our per-ceptions of the empirical world will always be in-fluenced by (often unconscious) a priori conceptionsof what reality is and ought to be.

The significance of this problem for a field likeeducation can be demonstrated by considering aterm such as learning. Few people would disagreewith the proposition that schools should promotelearning, but the term learning will mean quitedifferent things to different people, to Piagetianand Skinnerian psychologists, for instance. Beforea researcher can determine whether Program Aproduces more learning than Program B, the re-searcher must choose one of the paradigmsi.e.,one of the meaningsalluded to above or one of themultitude of other meanings which could be asso-ciated with the term learning. The meaning select-ed will influence the researcher's findings at leastas much as the empirical reality being described.

The situation is further complicated by the factthat, from certain paradigmatic perspectiVes, the

whole cause-effect way of thinking and talkingemployed by traditional researchers becomes prob-lematic. Freire (1970), Buber (1968), peace edu-cators like Galtung (1974), and a humanist readingof Dewey (see Kleibard 1975), for example, suggestthat educational practice should not be builtaround predetermined student learning outcomes,no matter what conception of learning the prede-termined outcomes reflect. This position suggeststhat rather than attempting to control students,teat:aers should engage in dialogue with students,and rather than transmitting a predefined curric-ulum to students, teachers should work with stu-dents to construct jointly the curriculum for theclass.

These prescriptive educational theories arecompatible with the theoretical descriptions ofhuman action put forth by symbolic interactionists,ethnomethodologists, and the 1980's version ofCronbach. Whether one agrees with these descrip-tive theories or not, they do provide an alternativeto Campbell and Stanley's conception of how thesocial world operates. As such, they remind us thatCampbell and Stanley's cause and effect conceptionof the social world is just that, an a priori concep-tion, a conception which is not determined by thefacts but rather determines what the facts are. Inshort, they reinforce a point made by Kant yearsago: It is impossible to talk of the nature of reali-ty with any sense of certainty because we can neverknow reality independent of the cognitive struc-tures which influence our perceptions.

Implit:ations

Ii this final section of the paper I want to brief-ly consider some implications of the two problemsoutlined above. General implications for rethink-ing the form and function of science will be dis-cussed. I will also provide more specific examplesof how the problems outlined above have influ-enced the development of the National Center forScience Teaching and Learning's research agenda.

IMPLICATIONS OF THE IDIOSYNCRACY PROBLEM. One

obvious implication of the problem of idiosyncra-cy is the realization that research will never be able

to provide prescriptions for practice; at best it willserve only a heuristic function. Of course, this doesnot mean that research is useless. Weiss (1982), forexample, suggests one heuristic function for socialscience research in her study of policy makers' uti-lization of research in formuiatini health policy.She indicates that research was of little use in prob-lem solving but very important in problem fram-ing. Social science research helped structurepolicy makers' thinking by supplying languagewith which to conceptualize policy questions. Inthe process, research directed policy makers atten-tion te possibilities and options which probablywould not have been considered in the absence ofresearch. Social science research can certainlyserve a similar role in the decision making of teach-ers and other educational professionals.

Social science research can also tell us what istypinl and give us some sense of what will typi-cally occur if we employ different types of educa-tional strategies. This sort of information will beespecially useful to educational policymakers.Unlike teachers who must be concerned with edu-cating idiosyncratic individuals, policy makers areprimarily concerned with aggregates. Like socialscientist, it is functional for policymakers to thinkin terms of types of people.

Policymakers, of course, must realize that ifeducators are to meet the needs of students theymust design policies and programs which allow forconsiderable discretion at the school and classroomlevels. After all, a probabalistic generalizationwhich tells us what will typically oecur ilso informsus that the atypical will occur with some individ-uals in some settings. Therefore even relativelydefinitive research findings do not automaticallytranslate into policies and programs. Once again,they serve only as a heuristic, not as a recipe.

The heuristic nature of research certainly sug-gests the need to reconsider standard operatingways of thinking about research ard standard op-erating ways of doing it. For instance, given Weiss'conclusion that research helps us to frame ratherthan to solve problems, it might make sense tothink of the whole matter of generalization in psy-

9

chological rather than in statistical and samplingterms. If we do this, single case studies suddenlyappear to have far more utility than was tradition-ally thought. (For an in-depth exploration of thisidea, see Donmoyer, 1990.)

At the very least, an understanding of the prob-lem of idiosyncracy forces us to lower our expec-tations for large scale research projects. At bestsuch projects will net findings which will have tobe shaped and adjusted by actors at the local levelto fit local contextual variables. This recognitionof the heuristic function of research and the im-portance of the local may lead to a blurring of thetraditional distinction between research and devel-opment. efforts, such as the work carried out byBerlir and White at the NCSTL (1992).

IMPLICATIONS OF THE IMMACULATE PERCEPTION

PROBLEM. Just as the idiosyncracy problem callsinto question traditional ideas aboutgeneralization, the problem of immaculateperception challenges traditional notions ofobjectivity. Furthermore, the recognition that theframes people bring to a problem to a large extentdictate the solutions they find also has influencedthe Center's decision to bring together diverseconstituencies (e.g. science teachers, teachereducators, administrators, practicing scientistsfrom multiple disciplines, business persons. andpolicymakers) which employ very different framesof reference to discuss and analyze the problemsof science education. The purpose in doing this isa bit different from the purpose which appears toundergird NSF's coalition building activities in thesystematic reform iMtiatives it has attempted topromote. We see such gatherings as a form ofinquiry, a form of inquiry which hearkens back toAristotelian notions of deliberation and practicalreasoning. (For further discussion of these ideas, seeDonmoyer, 1991a, 1991b, 1990.)

One additional implication of the problem ofimmaculate perception involves the need to expandthe substantive focus of our research. "We need tofocus on how teachers and others conceptualizescience and how different cultural perceptions canaffect the learning of science as it is conceptualized

10

within the culture of schools. To adequately hives-tigate cultural phenomena, we may also need toemploy methods from fields such as anthropology.Such methods are now widely accepted within theeducational research community in general but areless commonplace form in the subfield of scienceeducation.

Summary

IN THIS PAPER I REVIEWED TRADITIONAL NOTIONS

about the form and function of educational re-search, examined contemporary manifestations ofthese notions, and discussed two problems withtraditional ways of conceptualizing what researchis and what role research should play. I also brieflydiscussed some implications of these two problemsfor reconceptualizing the form and function ofeducational research in general and formulating aresearch agenda for the National Center of ScienceTeaching and Learning in particular.

12

References

Abt Associates. (1977). Education as experimentation: Aplanned variation model (Vol. !V A-))). Boston: Lbt Associates.

Berlin, D. F., & White, A. L (1992, April). Action research as asolution to the problem of knowledge utilization. Paper present-ed at the Annual Meeting of the American Educational ResearchAssociation, San Francisco, CA.

Blumer, H. (1969). Symbolic interactionism: Perspective andmethod. Englewood Cliffs, NJ: Prentice-HalL

Bobbitt, J. F. (1924). How to make a curriculum. Boston:Houghton Mifflin.

Buber, M. (1968). Education. Between man and man. NewYork: Macmillan.

Callahon, R. E. (1964). The cult of efficiency: A study of thesocial forces that have shaped the administration of publicschools. Chicago: University of Chicago Press.

Campbell, D., & Stanley, J. (1963). Experimental and quasi-experimental designs for research. Boston: Houghton Mifflin.

Clark, C., & Yinger, R. (1977). Research on teacher thinking.Curriculum Inquiry, 7, 279-309.

Cronbach, L. (1957). The two disciplines of sCientific psycholo-gy. American Psychologist, 12, 671-684.

Cronbach, L. (1975). Beyond the two disciplines of scientificpsychology. American Psychologist,1g, 116-127.

Cubberly, E. P. (1909). Changing conceptions of education.Boston: Houghton Mifflin.

Donmoyer, R. (1990). Curriculum evaluation and the negotia-tion of meaning. Languace Arts, 67, 274-286.

Donmoyer, R. (1991a). Administration and evaluation in post-positivist times. The Review Journal of Philosophy and SocialScience 16, 1-2.

Donmoyer, R. (1991b). Postpositivist evaluation: Give me a forinstance. Educational Administration Ouarterly, 27_, 265-296.

Fetterman, D. (1981). Blaming the victim. The problem ofevaluation design and federal involvement and reinforcingworld views of education. Human Organization AO, 67-77.

Freire, P. (1970). Pedagogy of the oppressed. New York: Sea-bury Press.

Galtung, J. (1974). On peace education. In C. Wulf (Ed.), Hand-book on Peace Education. Frankfurt/Main, FRG, and Oslo, Nor-way: International Peace Research Association.

1 .3

11

Garfinkle. H. (1967). Studies in ethnomethodolory. EnglewoodCliffs, NI: Prentice-Hall,

Good, R. (1991). Editorial. The Journal of Research in ScienceTeaching. g(4), 291-292.

Good, T., Biddle, B. J., & Brophy, J. E. (1975). Teachers make adifference. New York: Holt, Rinehart and Winston.

Haney, W., & Villaume, J. (1977, August/. The Follow throughplanned variation experiment Department of Health, Educat/on,and Welfare (Education Division), Office of Education, Office ofPlanning, Budgeting, and Evaluation, Washington, DC. (OEC-0:74-0394)

House, E. R., Glass, G. V, McLean, L D., & Walker, D. F. (1978).No simple answer: Critique of the follow-through evaluation.Educational Leadership,15, 462.

Kleibard, 11. (1975). Reappraisal: The Tyler rationale. In W.Pinar (Ed.), Curriculum theorizing: The reconceputalists. Berke-ley, CA: McCutchan.

Kuder, S. J. (1990). Effectiveness of the DISTAR reading pro-gram for children with learning disabilities. Journal of LearningMobilities 23 69-71.

Phillips, D. (1987). Philosophy, science, and social inquiry,. Ox-ford, UK: Pergamon Press.

Rivlin, A., & Tiinpane, M. (Eds.) (1975). Planned variationstudies. Washington, DC: Brookings Institution.

Shulman, L, & Lanier, J. (1977). Institute for research onteaching: An overview. Journal of Teacher Education, 28, 44-49.

Smith, M. (1991, April). Systemic reform. Paper presented atthe Annual Meeting of the American Educational Research Asso-ciation, Chicago.

Thorndike, E. L (1910). The contribution of psychology to edu-cation. The Journal of Educational Psycholou,1, 5-12.

Tyack, D. (1972). The one best system. Cambridge: HarvardUniversity Press.

Weiss, C. (1982). Policy research in the context of diffuse deci-sion making. ournal of Higher Education, a 619-639.

Wise, C. (1982). Policy research in the context of diffuse deci-sion making. Journal of Higher Education, 5_3, 619-639.